nutrients Article Network Analysis of the Potential Role of DNA Methylation in the Relationship between Plasma Carotenoids and Lipid Profile Bénédicte L. Tremblay 1,2, Frédéric Guénard 1,2 , Benoît Lamarche 1,2 , Louis Pérusse 1,3 and Marie-Claude Vohl 1,2,* 1 Institute of Nutrition and Functional Foods (INAF), Laval University, 2440 Hochelaga Blvd, Quebec City, QC G1V 0A6, Canada; [email protected] (B.L.T.); [email protected] (F.G.); [email protected] (B.L.); [email protected] (L.P.) 2 School of Nutrition, Laval University, 2425 rue de l’Agriculture, Quebec City, QC G1V 0A6, Canada 3 Department of Kinesiology, Laval University, 2300 rue de la Terrasse, Quebec City, QC G1V 0A6, Canada * Correspondence: [email protected]; Tel.: +1-418-656-2131 (ext. 404676) Received: 17 April 2019; Accepted: 31 May 2019; Published: 4 June 2019 Abstract: Variability in plasma carotenoids may be attributable to several factors including genetic variants and lipid profile. Until now, the impact of DNA methylation on this variability has not been widely studied. Weighted gene correlation network analysis (WGCNA) is a systems biology method used for finding gene clusters (modules) with highly correlated methylation levels and for relating them to phenotypic traits. The objective of the present study was to examine the role of DNA methylation in the relationship between plasma total carotenoid concentrations and lipid profile using WGCNA in 48 healthy subjects. Genome-wide DNA methylation levels of 20,687 out of 472,245 CpG sites in blood leukocytes were associated with total carotenoid concentrations. Using WGCNA, nine co-methylation modules were identified. A total of 2734 hub genes (17 unique top hub genes) were potentially related to lipid profile. This study provides evidence for the potential implications of gene co-methylation in the relationship between plasma carotenoids and lipid profile. Further studies and validation of the hub genes are needed. Keywords: carotenoids; DNA methylation; french canadians; hub genes; lipid profile; WGCNA 1. Introduction Cardiometabolic (CM) diseases comprise conditions ranging from insulin resistance and metabolic syndrome to cardiovascular disease and type-2 diabetes [1]. Healthy eating, including the consumption of fruits and vegetables, is associated with a favorable CM health [2]. Carotenoids, which are reliable biomarkers of fruit and vegetable intakes, are composed of hundreds of fat-soluble pigments [3]. However, six main carotenoids (α-carotene, β-carotene, β-cryptoxanthin, lutein, lycopene, and zeaxanthin) represent over 95% of total circulating carotenoids in human [4,5]. Variability among individuals in circulating carotenoids may be due to several factors, including age, sex, body weight, genetics, and lipid profile [3]. Lower circulating carotenoid concentrations are associated with lower plasma total cholesterol (TC), LDL-cholesterol (LDL-C), and HDL-cholesterol (HDL-C) concentrations [6]. Moreover, α- and β-carotene correlated with HDL-C and triglyceride (TG) concentrations [7], while β-crytoxanthin and zeaxanthin correlated with TG concentrations in a previous study by our group [8]. Thus, several studies have observed associations between circulating carotenoids and plasma lipid concentrations, which are both transported by lipoproteins [9]. Genetic variants have been shown to influence circulating carotenoid concentrations by causing differences in the absorption, assimilation, distribution, metabolism, and excretion of carotenoids [10–15]. Nutrients 2019, 11, 1265; doi:10.3390/nu11061265 www.mdpi.com/journal/nutrients Nutrients 2019, 11, 1265 2 of 14 Carotenoids and their derivatives (e.g., retinoid) have also been shown to modulate gene expression via several transcriptional systems [16]. However, very few studies have documented the relationship between carotenoids and DNA methylation, at least from a genome-wide viewpoint. In cell studies, lycopene had modest to null effects on DNA methylation of GSTP1, which is involved in prostate and breast cancers [17,18]. Moreover, a study of 165 overweight and obese subjects revealed an association between DNA methylation of HERV-w and TNFαin blood leukocytes and dietary intakes of β-carotene and carotenoids [19]. Interestingly, the reported protective effects of high fruit and vegetable intake on age-related diseases (coronary heart disease, stroke, type-2 diabetes) may be mediated by the association between blood carotenoids levels and extrinsic epigenetic age acceleration, which is a measure of epigenetic age [20]. Thus, circulating carotenoids and dietary intakes of carotenoids seem to impact DNA methylation. Moreover, there is more and more evidence that DNA methylation plays a role in the regulation of blood lipid levels and lipid metabolism-linked phenotypes and diseases [21]. No previous study has considered the involvement of genome-wide DNA methylation levels in the association between plasma carotenoids and lipid profile. Weighted gene correlation network analysis (WGCNA) is a widely used systems biology approach designed for high dimensional data (i.e. gene expression, DNA methylation, metabolites etc.) [22]. It allows finding gene clusters (modules) with highly correlated DNA methylation levels, relating these modules to phenotypic traits, and identifying key hub genes within modules that are related to phenotypic traits [22]. The objective of the present study was to examine the role of DNA methylation in the relationship between plasma carotenoid concentrations and lipid profile using WGCNA in 48 healthy subjects. First, cytosine-phosphate-guanine (CpG) sites whose DNA methylation levels are associated with carotenoid concentrations were identified using linear regressions. Second, WGCNA was used to identify specific modules and key hub genes related to lipid profile traits. The hypotheses were that genome-wide DNA methylation levels in blood leukocytes are associated with plasma total carotenoid concentrations and that clusters of genes associated with carotenoids are also correlated to lipid profile traits. Our results highlighted the potential implication of gene co-methylation in the relationship between plasma carotenoids and lipid profile. 2. Materials and Methods 2.1. Patients and Design The GENERATION Study aimed at evaluating familial resemblances in omics (DNA methylation [23] and gene expression [24]) and metabolic (metabolites [25] and carotenoids [8]) profiles in healthy families. The GENERATION Study comprises a total of 48 French-Canadian subjects from 16 families of Quebec City (Canada). Families composed of 16 mothers, six fathers, and 26 children lived under the same roof. Families comprised at least the mother and one child (8–18 years old). Parents had to be the biological parents of their child (or children), non-smokers, with body mass index (BMI) ranging between 18 and 35 kg/m2, and free of any metabolic conditions requiring treatment (Synthroid® (levothyroxine) and oral contraceptive were tolerated). Children also had to be non-smokers, in good general health and not using psycho-stimulators (Ritalin® (methylphenidate), Concerta® (methylphenidate), and Strattera® (atomoxetine)). Blood samples were taken from both parents and children at the Institute of Nutrition and Functional Foods (INAF). The experimental protocol was approved by the Ethics Committees of Laval University Hospital Research Center and Laval University. All participants signed an informed consent document. Parental consent was also obtained by signing the child consent document. 2.2. Anthropometric and CM Measurements Body weight and height were measured [26]. Blood samples were collected from an antecubital vein into vacutainer tubes containing EDTA after 12-hour fast and 48-hour alcohol abstinence. Plasma was separated by centrifugation (2500 g for 10 min at 4 C), and samples were aliquoted and frozen ( 80 C). ◦ − ◦ Plasma TC and TG concentrations were measured using enzymatic assays [27,28]. Precipitation of Nutrients 2019, 11, 1265 3 of 14 very-low density lipoprotein and LDL particles in the infranatant with heparin manganese chloride generated the HDL-C fraction for measurements of HDL levels [29]. LDL-C was estimated using the Friedewald formula [30]. The rocket immunoelectrophoretic method was used to measure plasma apolipoprotein B-100 (ApoB100) concentrations [31]. Plasma C-reactive protein (CRP) was measured by nephelometry using a sensitive assay (Prospec equipment Behring) [32]. Fasting plasma glucose concentrations were enzymatically measured [33]. Radioimmunoassay with polyethylene glycol separation was used to measure fasting plasma insulin concentrations [34]. 2.3. DNA Extraction and Methylation Analysis Genomic DNA was extracted from blood leukocytes using the GenElute Blood Genomic DNA kit (Sigma-Aldrich, St. Louis, MO, USA) in all 48 subjects. NanoDrop Spectrophotometer (Thermo Scientific, Wilmington, DE, USA) and PicoGreen DNA methods were used to quantify DNA. Infinium Human Methylation 450 array (Illumina, San Diego, CA, USA) was used to measure DNA methylation levels. McGill University and Genome Quebec Innovation Center (Montreal, QC, Canada) proceeded to the bisulfite conversion and quantitative DNA methylation analysis. Methylation data on all 485,577 CpG sites were analyzed using Illumina GenomeStudio software v2011.1 and the Methylation Module. All samples were retained after quality control steps [23]. GenomeStudio was used to perform global normalization